Hard to Fix

Intermittent errors and problems are very bad. It’s extremely difficult to pinpoint the cause of a problem if the problem doesn’t happen when you check. Like the classic example of a car that gives trouble, except when in the garage.

We had a call from a client today, with a machine that started to cause problems. During the call we identified a possible cause for the behaviour, and wanted the client to restart the machine, since the initialization behaviour will have signs that would allow us to verify (or disqualify) our idea.

The client guy calls back a few minutes later. He turned the machine off, turned it on, and everything is working perfectly. He can’t reproduce the problem.

The client did turn the machine off and on several times in the last day since the problem showed up, so that wasn’t it. But the behaviour shouldn’t have changed given the data we already had on it.

So now we can’t know if our idea was true. We don’t know what’s wrong with the machine. And we can’t try to fix and change anything since there’s no way to know if what we do makes a difference.

So far we just hope everything will keep working, since as long as it does everything is alright. But if the problem happens again, we have a serious… err… problem.

Problems are better when they’re just there until you kill them. This spontaneous solution thing is bad.

Leave a Reply

You must be logged in to post a comment.